Overview

DeepSeek Harness (DSH) adopts an “everything is a plugin” architecture, allowing developers to extend model inference capabilities through plugins. xby-pet-detect is a plugin focused on model inference and maintained by xby-skill. It is mainly used to address pet facial expression recognition, converting image input into structured emotion classification data.

Installation

Install it using DSH’s plugin management tool. It supports installation from a GitHub repository or local debugging.

Install from GitHub

Use the official installation command to add the plugin:

dsh plugin --profile web add github:xby_skill/xby-pet-detect

Local Development and Debugging

For local development, you can install it using an absolute path or debug it through a patch file:

dsh plugin --profile web add /absolute/path/to/xby-pet-detect

dsh web --profile web --patch /absolute/path/to/dsh-ocr-plugin/cordis.patch.yml

Core Features

The plugin provides three main inference interfaces, supporting different image input formats.

  1. set_xby_apikey
    Sets the API key in the chat. This action is automatically persisted and remains valid after restarting the DSH process.

  2. pet_emotion_recognition
    Recognizes pet facial expressions and outputs four emotion categories: Angry / Happy / Relaxed / Sad.
    - Input requirement: An image file link is required.

  3. pet_emotion_recognition_for_data_base64
    Recognizes pet facial expressions and outputs four emotion categories: Angry / Happy / Relaxed / Sad.
    - Input requirement: A BASE64-encoded image file is required.

  4. pet_emotion_recognition_for_data_file
    Recognizes pet facial expressions and outputs four emotion categories: Angry / Happy / Relaxed / Sad.
    - Input requirement: A file path to the image file is required.

Prerequisites and Configuration

The plugin depends on an external API service. Complete the following configuration before use.

Obtain an API Key

You need to register on the Xiaobenyang official website and obtain an API key.
- Website: https://xiaobenyang.com

Usage Steps

  1. In the DSH chat environment, call the set_xby_apikey command to set the key.
  2. Select the corresponding recognition interface based on the image source:
    - If the image is a link, use pet_emotion_recognition.
    - If the image is a Base64 string, use pet_emotion_recognition_for_data_base64.
    - If the image is a local file, use pet_emotion_recognition_for_data_file.

Ecosystem Background

The DSH plugin directory (https://www.skillhub.cn/plugins/xby-skill/xby-pet-detect) has no official affiliation with DeepSeek or High-Flyer (Huanfang) Research Institute. The plugin is released under the MIT License, and developers are free to use and modify it.